Use GlycanDock in a blind docking like setting

[zillmerhanne]

I would like to use GlycanDock to perform glycan docking (in a blind docking like setting). Thanks to the great supplementary information in the paper, I already managed to obtain a putative protein-glycoligand complex as input structure. My glycoligand of interest is a 1,3-D-Glucan trimer (saccharide sequence: a-D-Glcp-(1->3)-a-D-Glcp-(1->3)-a-D-Glcp). When creating the initial conformer for the 1,3-D-Glucan-trimer, each monomer is assigned with a unique residue number.
However, this residue numbering is lost when using molfile_to_params.py to convert the .sdf file into a RosettaLigand-compatible .pdb file and also not found in the output structure of the initial docking using RosettaLigand. Instead all of the glycoligands are assigned to residue number 1. When using this putative complex as input structure for GlycanDock, I get the following warning and some of the atoms of the glycoligand are removed:

core.io.pose_from_sfr.PoseFromSFRBuilder: [ WARNING ] discarding 46 atoms at position 226 in file prot_GLC_complex.pdb. Best match rsd_type: ->3)-alpha-D-Glcp:reducing_end:non-reducing_end:2-AcNH

I used GLC as ligand name and added the -alternate_3_letter_codes pdb_sugar flag for the GlycanDock run.
Using an experimental structure as input, e.g., PDB ID: 2k46 (contains 1,3-D-glucan dimer with different residues assigned to each monomer), GlycanDock runs without a problem.
Is there any way to avoid this behavior or is there any additional step that I overlooked?